Over the past century, understanding of fracture has been significantly advanced due to many successful developments of experimental, theoretical, and computational methods. In particular, with the advent of high-performance computers, atomic-level molecular dynamics (MD) simulation has become a powerful tool to study material fracture at a very high resolution. Nevertheless, the current state-of-the-art MD simulations can handle several hundred billion atoms, amounting to a volume less than a cubic micron. It is unlikely, in the near future, for MD to fully resolve the complex fracture problems, such as arbitrary crack branching, crack-tip blunting, and brittle-to-ductile (BTD) transition, with realistic material size. To understand the failure of materials from the atomic to the macroscopic level, a coarse-grained (CG) atomistic model as well as an adaptive concurrent atomistic-continuum (CAC) method was developed to predict the dynamic brittle and ductile fracture in a variety of materials.
