Efficient autoencoder pipeline for discovering high entropy alloys with molecular dynamics data

In this work, we utilize computationally efficient molecular dynamics simulations to create a machine learning pipeline for discovery of crystalline multi-component alloys. We employ high-quality interatomic potentials to create a dataset of NiFeCr structures and apply crystal diffusion variational...

Full description

Saved in:
Bibliographic Details
Main Authors: Amirhossein D Naghdi, Grzegorz Kaszuba, Stefanos Papanikolaou, Andrzej Jaszkiewicz, Piotr Sankowski
Format: Article
Language:English
Published: IOP Publishing 2025-01-01
Series:Machine Learning: Science and Technology
Subjects:
Online Access:https://doi.org/10.1088/2632-2153/addf0f
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849723372314820608
author Amirhossein D Naghdi
Grzegorz Kaszuba
Stefanos Papanikolaou
Andrzej Jaszkiewicz
Piotr Sankowski
author_facet Amirhossein D Naghdi
Grzegorz Kaszuba
Stefanos Papanikolaou
Andrzej Jaszkiewicz
Piotr Sankowski
author_sort Amirhossein D Naghdi
collection DOAJ
description In this work, we utilize computationally efficient molecular dynamics simulations to create a machine learning pipeline for discovery of crystalline multi-component alloys. We employ high-quality interatomic potentials to create a dataset of NiFeCr structures and apply crystal diffusion variational autoencoder to maximize their mechanical properties, i.e. bulk modulus. As part of the experiment, we utilize local search coupled with classical interatomic potentials to explore the local structure space and show that utilization of this procedure greatly improves optimization capability of the neural model. We also expand the model with an extra submodule, which attains 42% improvement on modeling the crystalline phase of the structures. Ultimately, we verify the global stability of the created structures with quantum mechanical calculation methods.
format Article
id doaj-art-3b0e49ec01de4718a5a4ebe76300cd8b
institution DOAJ
issn 2632-2153
language English
publishDate 2025-01-01
publisher IOP Publishing
record_format Article
series Machine Learning: Science and Technology
spelling doaj-art-3b0e49ec01de4718a5a4ebe76300cd8b2025-08-20T03:11:03ZengIOP PublishingMachine Learning: Science and Technology2632-21532025-01-016202505510.1088/2632-2153/addf0fEfficient autoencoder pipeline for discovering high entropy alloys with molecular dynamics dataAmirhossein D Naghdi0https://orcid.org/0000-0002-2709-8111Grzegorz Kaszuba1https://orcid.org/0000-0002-8993-1703Stefanos Papanikolaou2https://orcid.org/0000-0001-5239-1275Andrzej Jaszkiewicz3Piotr Sankowski4NOMATEN Centre of Excellence , Warsaw, PolandIDEAS NCBR , Warsaw, Poland; Poznań University of Technology , Poznań, PolandNOMATEN Centre of Excellence , Warsaw, PolandPoznań University of Technology , Poznań, PolandUniversity of Warsaw , Warsaw, Poland; MIM Solutions , Warsaw, PolandIn this work, we utilize computationally efficient molecular dynamics simulations to create a machine learning pipeline for discovery of crystalline multi-component alloys. We employ high-quality interatomic potentials to create a dataset of NiFeCr structures and apply crystal diffusion variational autoencoder to maximize their mechanical properties, i.e. bulk modulus. As part of the experiment, we utilize local search coupled with classical interatomic potentials to explore the local structure space and show that utilization of this procedure greatly improves optimization capability of the neural model. We also expand the model with an extra submodule, which attains 42% improvement on modeling the crystalline phase of the structures. Ultimately, we verify the global stability of the created structures with quantum mechanical calculation methods.https://doi.org/10.1088/2632-2153/addf0fgenerative AIhigh entropy alloyscomposition optimizationmolecular dynamics
spellingShingle Amirhossein D Naghdi
Grzegorz Kaszuba
Stefanos Papanikolaou
Andrzej Jaszkiewicz
Piotr Sankowski
Efficient autoencoder pipeline for discovering high entropy alloys with molecular dynamics data
Machine Learning: Science and Technology
generative AI
high entropy alloys
composition optimization
molecular dynamics
title Efficient autoencoder pipeline for discovering high entropy alloys with molecular dynamics data
title_full Efficient autoencoder pipeline for discovering high entropy alloys with molecular dynamics data
title_fullStr Efficient autoencoder pipeline for discovering high entropy alloys with molecular dynamics data
title_full_unstemmed Efficient autoencoder pipeline for discovering high entropy alloys with molecular dynamics data
title_short Efficient autoencoder pipeline for discovering high entropy alloys with molecular dynamics data
title_sort efficient autoencoder pipeline for discovering high entropy alloys with molecular dynamics data
topic generative AI
high entropy alloys
composition optimization
molecular dynamics
url https://doi.org/10.1088/2632-2153/addf0f
work_keys_str_mv AT amirhosseindnaghdi efficientautoencoderpipelinefordiscoveringhighentropyalloyswithmoleculardynamicsdata
AT grzegorzkaszuba efficientautoencoderpipelinefordiscoveringhighentropyalloyswithmoleculardynamicsdata
AT stefanospapanikolaou efficientautoencoderpipelinefordiscoveringhighentropyalloyswithmoleculardynamicsdata
AT andrzejjaszkiewicz efficientautoencoderpipelinefordiscoveringhighentropyalloyswithmoleculardynamicsdata
AT piotrsankowski efficientautoencoderpipelinefordiscoveringhighentropyalloyswithmoleculardynamicsdata