Efficient training of machine learning potentials for metallic glasses: CuZrAl validation
Antoni Wadowski, Anshul D.S. Parmar, Jesper Byggmästar, Jan S. Wróbel, Mikko J. Alava, Silvia Bonfanti · 2025
Read Efficient training of machine learning potentials for metallic glasses: CuZrAl validation by Antoni Wadowski, Anshul D.S. Parmar, Jesper Byggmästar, Jan S. Wróbel, Mikko J. Alava, Silvia Bonfanti (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-00589.txt.
Introduction
Antoni Wadowski and co-authors offer a cheaper way to train machine-learned atom models in a paper whose head dates it to 2024. Such models let scientists simulate metallic glasses, metals with a disordered structure. Their disorder makes good training data scarce and costly.
The team starts from a simple surrogate model and swaps atoms with a Monte Carlo method to build glassy states across a huge cooling range. A few exact quantum calculations then correct the set, avoiding mass data production. Tested on a copper-zirconium-aluminum glass, the model matched lab data and classic model results for structure, motion, energy, and strength. The path should scale to richer alloys.
Catalog record
| CREATOR | Antoni Wadowski, Anshul D.S. Parmar, Jesper Byggmästar, Jan S. Wróbel, Mikko J. Alava, Silvia Bonfanti |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 30 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-00589
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Efficient training of machine learning potentials for metallic glasses: CuZrAl validation Antoni Wadowski, Anshul D.S. Parmar, Jesper Byggmästar, Jan S. Wróbel, Mikko J. Alava, Silvia Bonfanti Abstract Interatomic potentials play a vital role in revealing microscopic details and structure-property relations, which are fundamental for multiscale simulations and to assist high-throughput experiments. For metallic glasses, developing these potentials is challenging due to the complexity of their unique disordered structure. As a result, chemistry-specific interaction potentials for this important class of materials are often missing. Here, we solve this gap by implementing an efficient methodology for designing machine learning interatomic potentials (MLIPs) for metallic glasses, and we benchmark it with the widely studied CuZrAl system. By combining a Lennard-Jones surrogate model with swap-Monte Carlo sampling and Density Functional Theory (DFT) corrections, we capture diverse amorphous structures from 14 decades of supercooling. These distinct structures provide robust and efficient training of the model and applicability to the wider spectrum of energies. This approach reduces the need for extensive DFT and ab initio optimization datasets, while maintaining high accuracy. Our MLIP shows results comparable to the classical Embedded Atom Method (EAM) available for CuZrAl, in predicting structural, energetic, and mechanical properties. This work paves the way for the development of new MLIPs for complex metallic glasses, including emerging multicomponent and high entropy metallic glasses. I Introduction Metallic glasses (MGs) are an extraordinary class of materials composed of metallic elements arranged in a disordered atomic structure. This unique structure gives them a range of exceptional properties, such as high strength, hardness, and elasticity Greer 1995 ; Kruzic 2016 ; Suryanarayana and Inoue 2017 . Consequently, MGs are increasingly being used in many different fields, for example, electronics, biomedical engineering, nanotechnology, and aerospace Gao et al. 2022 ; Greer et al. 2023 ; Sohrabi et al. 2024 . However, the disordered nature of MGs is also a limitation, as it gives rise to a complex and rugged potential energy landscape (PEL) Debenedetti and Stillinger 2001 ; Bonfanti and Kob 2017 further complicated by the vast compositional variability Forrest and Greer 2023 . The composition space of MGs is, in fact, mainly unexplored, as the number of possible combinations of metallic elements and their proportions is extremely large. Predicting their properties and designing optimal compositions is therefore a challenging task. For this reason, the discovery of novel MGs has traditionally relied on intensive experimental trials and errors Telford 2004 , only recently supplemented with machine learning methods Forrest and Greer 2023 ; Merchant et al. 2023 ; Makinen et al. 2024 , combined with high-throughput experimentation Sarker et al. 2022 ; Wu et al. 2024 . To efficiently explore atomic-scale structures, in silico calculations have become fundamental, offering microscopic insights that are often inaccessible due to experimental limitations Binder and Kob 2011 . In particular, the development of accurate interatomic potentials that mimic atomic interactions is central to these computer simulations. However, chemistry-specific potentials for MGs are often missing, due to their disordered structure and composition complexity. As an alternative route to precisely describe atomic interactions, density functional theory (DFT) and ab initio calculations can be considered. Nevertheless, their computational cost restricts applicability to small systems and timescales, and cannot capture the rugged energy landscape of MGs. On the other hand, simplified potentials used to describe MGs, such as binary mixture Lennard-Jones (LJ) potentials Kob and Andersen 1995 or polydisperse systems Ninarello et al. 2017 , are not meant to account for specific chemical compositions. Therefore, they do not explain, for instance, why “substitutional metallic glasses”, where one metal with a similar radius replaces another, show different dynamical properties, as seen in experimental reality Voigt et al. 2024 . Finally, classical potentials for metallic elements, such as embedded atom method (EAM), tuned semi-empirical, are computationally efficient even for bulk systems. However, EAM potentials are available only for specific compositions and present challenges in parameterization that reduce their transferability for the study of realistic, multicomponent systems Daw et al. 1993 ; Müser et al. 2023 . These limitations led to the development of machine learning interatomic potentials (MLIPs), which allow to approximate the PEL with near-DFT accuracy, while enabling large-scale simulations Behler 2016 ; Deringer et al. 2019 . MLIPs have been successfully employed for a range of disordered systems Bartók et al. 2018 ; Deringer and Csányi 2017 ; Sosso et al. 2012 ; Sosso et al. 2012 ; Xie et al. 2021 ; Zhao et al. 2023 ; Bertani et al. 2024 . However, MLIPs have also some drawbacks: i) Their accuracy and robustness rely heavily on the quality of the training data, which for glasses is often limited to time scales much shorter than those observed experimentally. ii) The need for large datasets and high dimensionality increases computational complexity, raising challenges for their transferability and overall robustness Montes de Oca Zapiain et al. 2022 ; Qi et al. 2024 ; Wang et al. 2024 . To fill these gaps, this work presents an efficient methodology for tailoring MLIPs to MGs, combining a computationally inexpensive LJ-surrogate model, accelerated swap-Monte Carlo sampling, and DFT corrections, to obtain distinct amorphous structures corresponding to timescales comparable to experiments, achieving physical accuracy and computational efficiency. First, a LJ-surrogate model parametrized with DFT Caro et al. 2015 , provides a simple and effective framework to explore the rugged PEL of MGs. To extend the range of disordered configurations and access deeply supercooled states, which are otherwise unattainable with conventional simulation methods, non-local moves using swap-Monte Carlo sampling are performed Ninarello et al. 2017 ; Parmar et al. 2020 . Finally, single-point DFT corrections are applied to the obtained structures to refine energies and forces with first-principles, capturing realistic chemistry-specific interactions and generating high-accuracy data for training-testing the MLIP. This hybrid approach bypasses the most computationally expensive aspects of MLIP development for MGs. Realistic amorphous configurations are generated through an accelerated sampling of the effective LJ-surrogate PEL, and single-point DFT corrections refine these structures, eliminating the need for full DFT optimizations. These steps address the challenges of dataset quality and computational cost, resulting in transferable MLIP performance for modeling complex MGs. Figure 1: Schematic of PEL exploration of the LJ-surrogate model with swap-MC, used for accelerated sampling and connection to DFT. Swap-MC simulations provide samples from the extended regime of the LJ-surrogate PEL (black line), which are both associated with conventional simulation methods (grey region) and the deeper energy minima from the astronomical time scales (blue region). Configurations obtained with our LJ-surrogate model (yellow particles as representative) can be directly utilized to investigate the DFT PEL (red line) trough single-point DFT corrections. The methodology bypasses the necessity of expensive DFT optimizations and enables the exploration of the largest range of supercooling. To demonstrate the applicability of the proposed methodology, we employ machine learning neuroevolution potentials (NEP) Fan et al. 2022 ; Song et al. 2024 and design a new MLIP for CuZrAl MG Inoue and Zhang 2002 ; Das et al. 2005 ; Cheung and Shek 2007 ; Yu and Bai 2008 ; Pauly et al. 2010 ; Poltronieri et al. 2023 ; Alvarez-Donado et al. 2023 ; Mäkinen et al. 2024 ; Jiang et al. 2024 . The effectiveness of the