Titre :

Machine learning surrogate models for prediction of point defect vibrational entropy

Auteur(s) :

Lapointe, Clovis [Auteur]
Département des Matériaux pour le Nucléaire [DMN]
Swinburne, Thomas D. [Auteur]
Centre National de la Recherche Scientifique [CNRS]
Centre Interdisciplinaire de Nanoscience de Marseille [CINaM]
Thiry, Louis [Auteur]
Mallat, Stéphane [Auteur]
Centre de Mathématiques Appliquées [CMAP]
Collège de France - Chaire Sciences des données
Proville, Laurent [Auteur]
Becquart, Charlotte [Auteur]
Unité Matériaux et Transformations - UMR 8207 [UMET]
Marinica, Mihai-Cosmin [Auteur]

Titre de la revue :

Physical Review Materials

Nom court de la revue :

Phys. Rev. Materials

Numéro :

4

Éditeur :

American Physical Society (APS)

Date de publication :

2020-06-15

ISSN :

2475-9953

Discipline(s) HAL :

Chimie/Matériaux
Physique [physics]/Matière Condensée [cond-mat]/Science des matériaux [cond-mat.mtrl-sci]

Résumé en anglais : [en]

The temperature variation of the defect densities in a crystal depends on vibrational entropy. This contribution to the system thermodynamics remains computationally challenging as it requires a diagonalization of the ...
Lire la suite >The temperature variation of the defect densities in a crystal depends on vibrational entropy. This contribution to the system thermodynamics remains computationally challenging as it requires a diagonalization of the system's Hessian which scales as O(N3) for a crystal made of N atoms. Here, to circumvent such a heavy computational task and make it feasible even for systems containing millions of atoms, the harmonic vibrational entropy of point defects is estimated directly from the relaxed atomic positions through a linear-in-descriptor machine learning approach of order O(N). With a size-independent descriptor dimension and fixed model parameters, an excellent predictive power is demonstrated on a wide range of defect configurations, supercell sizes, and external deformations well outside the training database. In particular, formation entropies in a range of 250kB are predicted with less than 1.6kB error from a training database whose formation entropies span only 25kB (training error less than 1.0kB). This exceptional transferability is found to hold even when the training is limited to a low-energy superbasin in the phase space while the tests are performed for a different liquid-like superbasin at higher energies.Lire moins >

Langue :

Anglais

Comité de lecture :

Oui

Audience :

Internationale

Établissement(s) :

Université de Lille
CNRS
INRA
ENSCL

Collections :

• Unité Matériaux et Transformations (UMET) - UMR 8207

Équipe(s) de recherche :

Métallurgie Physique et Génie des Matériaux

Date de dépôt :

2020-11-02T15:37:43Z
2020-11-07T21:03:52Z
2020-11-16T10:38:44Z
2020-11-25T15:36:41Z