Under external mechanical loading, glassy materials, ranging from soft matter systems to metallic alloys, often respond via formation of inhomogeneous flow patterns, during yielding. These inhomogeneities can be precursors to catastrophic failure, implying that a better understanding of their underlying mechanisms could lead to the design of smarter materials. Here, extensive molecular dynamics simulations are used to reveal the emergence of heterogeneous dynamics in a binary Lennard-Jones glass, subjected to a constant strain rate. At a critical strain, this system exhibits for all considered strain rates a transition towards the formation of a percolating cluster of mobile regions. We give evidence that this transition belongs to the universality class of directed percolation. Only at low shear rates, the percolating cluster subsequently evolves into a transient (but long-lived) shear band with a diffusive growth of its width. Finally, the steady state with a homogeneous flow pattern is reached. In the steady state, percolation transitions also do occur constantly, albeit over smaller strain intervals, to maintain the stationary plastic flow in the system.

Molecular dynamics computer simulations of a binary Lennard–Jones glass under shear are presented. The mechanical response of glassy states having different thermal histories is investigated by imposing a wide range of external shear rates, at different temperatures. The stress-strain relations exhibit an overshoot at a strain of around 0.1, marking the yielding of the glass sample and the onset of plastic flow. The amplitude of the overshoot shows a logarithmic behavior with respect to a dimensionless variable, given by the age of the sample times the shear rate. Dynamical heterogeneities having finite lifetimes, in the form of shear bands, are observed as the glass deforms under shear. By quantifying the spatial fluctuations of particle mobility, we demonstrate that such shearbanding occurs only under specific combinations of imposed shear-rate, age of glass, and ambient temperature.

We use a computationally efficient graph-cut method (GCM) to obtain the ground-state morphologies (at zero temperature) of the random-field Ising model in $d=2,3$. The GCM enables us to obtain comprehensive numerical results on large-scale systems. We analyze the morphologies by computing correlation functions and structure factors. These quantities enable us to precisely evaluate characteristic properties, e.g., domain sizes, scaling functions, roughness exponents, fractal dimensions, etc.