Mesoscale modelling of plastic instabilities using machine learning
This project aims to study the predictability of deformations in small crystals using machine learning models with data from phase field crystal (PFC) model for plasticity.
Prediction and control of mechanical deformations are central concepts in the design of advanced materials. Yet, submicron materials reveal an erratic response to applied stress. The global stress-strain response curves vary from sample to sample and for different sample sizes. For crystalline materials, the wild fluctuations in the stress-strain response curves are attribute to heterogeneous, time-dependent distributions of crystal defects, such as dislocations. Modelling the evolution of dislocation ensembles is key to understanding their collective effects on the yield stress, plastic instabilities and their finite-size effects.
This project aims to study the predictability of deformations in small crystals using machine learning models with data from phase field crystal (PFC) model for plasticity. We use the newly proposed PFC model as a versatile minimal model for dislocation dynamics that has no direct control parameters for kinetic rates or interaction potential.
The candidate will develop an efficient numerical implementation of the PFC model suitable for large-scale simulations of dislocation ensembles in 2D and 3D and under different boundary conditions. The candidate will apply machine learning techniques such as regression neural networks and support vector machine to analyze the deformation predictability depending on key factor such as the stress and crystal size.
- MSc in physics, preferably in statistical, computational or condensed matter physics.
- Candidates with documented experience in scientific programming (solving non-linear PDE's, data analysis, finite element methods) or theoretical background in statistical physics will be prioritized.
Call 2: Project start autumn 2022
This project is in call 2, starting autumn 2022.