Demand-based coupling of the scales in concurrent atomistic-to-continuum models at finite temperature

Concurrent atomistic-to-continuum multiscale models commonly couple the coarse and fine scale in a fixed time interval fashion. This coupling is generally not the most efficient, as the coarse scale solution is also computed in times when its boundary conditions do not change significantly. Using a prototypical representative of these models, namely the coupled atomistic and discrete dislocation (CADD) method, we show that substantial computational savings are possible by using a demand-based coupling of the scales instead. To this end, we present an algorithm that uses digital low-pass filtering and demonstrate its capability in three numerical examples.