Universität Stuttgart

Permanent URI for this communityhttps://elib.uni-stuttgart.de/handle/11682/1

Browse

Subject: search.filters.subject.530Institute: search.filters.UBSInstitut.Institut für MaterialwissenschaftInstitute: search.filters.UBSInstitut.Fakultätsübergreifend / Sonstige EinrichtungAuthor: search.filters.author.Gubaev, Konstantin

Search Results

Now showing 1 - 2 of 2
  • Thumbnail Image
    ItemOpen Access
    Finite-temperature interplay of structural stability, chemical complexity, and elastic properties of bcc multicomponent alloys from ab initio trained machine-learning potentials
    (2021) Gubaev, Konstantin; Ikeda, Yuji; Tasnádi, Ferenc; Neugebauer, Jörg; Shapeev, Alexander V.; Grabowski, Blazej; Körmann, Fritz
    An active learning approach to train machine-learning interatomic potentials (moment tensor potentials) for multicomponent alloys to ab initio data is presented. Employing this approach, the disordered body-centered cubic (bcc) TiZrHfTax system with varying Ta concentration is investigated via molecular dynamics simulations. Our results show a strong interplay between elastic properties and the structural ω phase stability, strongly affecting the mechanical properties. Based on these insights we systematically screen composition space for regimes where elastic constants show little or no temperature dependence (elinvar effect).
  • Thumbnail Image
    ItemOpen Access
    Performance of two complementary machine-learned potentials in modelling chemically complex systems
    (2023) Gubaev, Konstantin; Zaverkin, Viktor; Srinivasan, Prashanth; Duff, Andrew Ian; Kästner, Johannes; Grabowski, Blazej
    Chemically complex multicomponent alloys possess exceptional properties derived from an inexhaustible compositional space. The complexity however makes interatomic potential development challenging. We explore two complementary machine-learned potentials - the moment tensor potential (MTP) and the Gaussian moment neural network (GM-NN) - in simultaneously describing configurational and vibrational degrees of freedom in the Ta-V-Cr-W alloy family. Both models are equally accurate with excellent performance evaluated against density-functional-theory. They achieve root-mean-square-errors (RMSEs) in energies of less than a few meV/atom across 0 K ordered and high-temperature disordered configurations included in the training. Even for compositions not in training, relative energy RMSEs at high temperatures are within a few meV/atom. High-temperature molecular dynamics forces have similarly small RMSEs of about 0.15 eV/Å for the disordered quaternary included in, and ternaries not part of training. MTPs achieve faster convergence with training size; GM-NNs are faster in execution. Active learning is partially beneficial and should be complemented with conventional human-based training set generation.