Deep learning; Image generation and processing; LSTM; PCA; Python development; Software
Résumé :
[en] The passage describes the development of a numerical framework in Python to create and process a large dataset for time-series prediction using Deep Learning algorithms. The dataset is generated by solving the Cahn–Hilliard equation for spinodal decomposition of a binary alloy and is labeled to train the algorithms. Prior to training, dimensionality reduction is performed using Auto-encoders and Principal Component Analysis. The framework identifies three distinct latent dimensions/spaces for the datasets. The primary dataset was generated by running up to 10,000 High-Fidelity Phase-Field simulations in parallel using High-Performance Computing (HPC). The framework is compatible with all major operating systems and has been thoroughly tested on Python 3.7 and later versions.
Centre/Unité de recherche :
UEE - Urban and Environmental Engineering - ULiège
Disciplines :
Sciences informatiques
Auteur, co-auteur :
Fetni, Seifallah ; UEE Research Unit, University of Liège, Belgium
Delahaye, Jocelyn ; Université de Liège - ULiège > Aérospatiale et Mécanique (A&M)
Habraken, Anne ; Université de Liège - ULiège > Département ArGEnCo > Département Argenco : Secteur MS2F
Langue du document :
Anglais
Titre :
Python Data Driven framework for acceleration of Phase-Field simulations[Formula presented]
Date de publication/diffusion :
septembre 2023
Titre du périodique :
Software Impacts
eISSN :
2665-9638
Maison d'édition :
Elsevier
Volume/Tome :
17
Pagination :
100563
Peer reviewed :
Peer reviewed vérifié par ORBi
Tags :
CÉCI : Consortium des Équipements de Calcul Intensif
Uliege FSA Faculty Research grant and CECI. Computational resources have been provided by the Consortium des Équipements de Calcul Intensif (CÉCI), funded by the Fonds de la Recherche Scientifique de Belgique (F.R.S.-FNRS), Belgium under Grant No. 2.5020.11 and by the Walloon Region, Belgium . A special thank to Mr. David Colignon for his availability and great support to successfully achieve computational tasks.
Hu, C., Martin, S., Dingreville, R., Accelerating phase-field predictions via recurrent neural networks learning the microstructure evolution in latent space. Comput. Methods Appl. Mech. Engrg., 397, 2022, 115128, 10.1016/j.cma.2022.115128.
Choudhary, K., et al. Recent advances and applications of deep learning methods in materials science. npj Comput. Mater., 8(1), 2022, 59, 10.1038/s41524-022-00734-6.
Teurtrie, A., et al. Espm: A python library for the simulation of STEM-EDXS datasets. Ultramicroscopy, 249, 2023, 113719, 10.1016/j.ultramic.2023.113719.
Toriyama, M.Y., Qu, J., Gomes, L.C., Ertekin, E., VTAnDeM: A python toolkit for simultaneously visualizing phase stability, defect energetics, and carrier concentrations of materials. Comput. Phys. Comm., 287, 2023, 108691, 10.1016/j.cpc.2023.108691.
S. Fetni, J. Delahaye, L. Duchêne, A. Mertens, A.M. Habraken, Adaptive time stepping approach forPhase-Field modeling of phase separation and precipitates coarsening in additive manufacturing alloys - COMPLAS 2021, in: COMPLAS 2021-16th Int. Conf. Comput. Plast. Fundam. Appl, 2021, pp. 1–12, http://dx.doi.org/10.23967/complas.2021.009.
Fetni, S., et al. Capabilities of auto-encoders and principal component analysis of the reduction of microstructural images; application on the acceleration of phase-field simulations. Comput. Mater. Sci., 216, 2023, 111820, 10.1016/j.commatsci.2022.111820.
Montes de Oca Zapiain, D., Stewart, J.A., Dingreville, R., Accelerating phase-field-based microstructure evolution predictions via surrogate models trained by machine learning methods. npj Comput. Mater., 7(1), 2021, 10.1038/s41524-020-00471-8.
