Machine learning plasticity and dislocations: avalanches and predictability

9 April, 2019
h. 14.00
Room Caldirola
Physics Department

Mikko Alava
Aalto University, Finland

The yielding and avalanches in crystal plasticity do not adhere to the picture of classical (depinning) non-equilibrium transition, and I will discuss the how and why in this talk. In two dimensional discrete dislocation plasticity I will discuss how the use of machine learning as a tour de force shows how the disorder or stress landscape in fact is relevant, and how that allows one to predict the coarse-grained behavior of individual systems or samples to a large degree, in contrast to what is usually expected of critical or avalanching systems. (Talk based on Salmenjoki et al. Nature Comm. December 2018). In collaboration with Lasse Laurson (Tampere University, Finland), Henri Salmenjoki (Aalto, Finland)

published on 3/28/2019