Parallel Scientific Computing with Applications in Material Science and Metallurgy

A new software tool for the solution of complex time-dependent systems of partial differential is presented. High levels of parallelization are achieved in a framework that allows the developer to implement ad-hoc solvers for computationally challenging problems on a higher abstraction level without the need to understand in the low-level parallel implementation. Moreover, thanks to this new implementation, advanced numerical methods, such as mesh adaptivity, implicit time stepping, and multigrid methods can be employed with ease. Here the implementation of this new tool is presented and validated against simple elliptic and more complex phase-field models. Its parallel performance is then assessed.