Machine learning plastic deformation of crystals.

摘要: Plastic deformation of micron-scale crystalline solids exhibits stress-strain curves with significant sample-to-sample variations. It is a pertinent question if this variability purely random or to some extent predictable. Here we show, by employing machine learning techniques such as regression neural networks and support vector machines that predictability evolves strain crystal size. Using data from discrete dislocations dynamics simulations, the models are trained infer mapping features pre-existing dislocation configuration curves. The vs relation non-monotonic system size effect: larger systems more Stochastic avalanches give rise fundamental limits for intermediate strains. However, large-strain samples can be predicted surprisingly well.