Accelerating simulations of strained-film growth by deep learning: Finite element method accuracy over long time scales
Open Access
- 29 August 2024
- journal article
- Published by AIP Publishing in APL Machine Learning
- Vol. 2 (3)
Abstract
A convolutional neural network is trained on a large dataset of suitably randomized film profiles and corresponding elastic energy densities ρɛ, computed by the finite element method. The trained model provides quantitative predictions of ρɛ for arbitrary profiles, surrogating its explicit calculation, and is used for the time integration of partial differential equations describing the evolution of strained films. The close match found between the neural network predictions and the “ground-truth” evolutions obtained by the finite element method calculation of ρɛ, even after tens-of-thousands of integration time-steps, validates the approach. A substantial computational speed up without significant loss of accuracy is demonstrated, allowing for million-steps simulations of islands growth and coarsening. The intriguing possibility of extending the domain size is also discussed.Keywords
Funding Information
- Ministerio de Ciencia e Innovación (MCIN/AEI/ 10.13039/501100011033 under Project No. PID2020-115118GB-I00)
- ICSC-Centro Nazionale di Ricerca in High Performance Computing, Big Data and Quantum Computing
This publication has 49 references indexed in Scilit:
- Gaussian process regression as a surrogate model for the computation of dispersion relationsComputer Methods in Applied Mechanics and Engineering, 2024
- Predicting dynamic responses of continuous deformable bodies:A graph-based learning approachComputer Methods in Applied Mechanics and Engineering, 2023
- A framework for neural network based constitutive modelling of inelastic materialsComputer Methods in Applied Mechanics and Engineering, 2023
- Automated Analysis of Continuum Fields from Atomistic Simulations Using Statistical Machine LearningAdvanced Engineering Materials, 2022
- Machine learning materials physics: Multi-resolution neural networks learn the free energy and nonlinear elastic response of evolving microstructuresComputer Methods in Applied Mechanics and Engineering, 2020
- Anisotropy effect on strain-induced instability during growth of heteroepitaxial filmsJournal of Materials Science, 2017
- Equilibrium surface roughness of a strained epitaxial film due to surface diffusion induced by interface misfit dislocationsMechanics of Materials, 1995
- The stress driven instability in elastic crystals: Mathematical models and physical manifestationsJournal of Nonlinear Science, 1993
- On the stability of surfaces of stressed solidsActa Metallurgica, 1989
- Interface morphology development during stress corrosion cracking: Part I. Via surface diffusionMetallurgical Transactions, 1972