Multiscale coupling methods constitute a powerful and efficient tool to study local phenomena that occur at the atomic scale. With these approaches, one may couple an atomic description, which models the material at the finest scale – such as molecular dynamics – with a macroscopic model (e.g. continuum mechanics). The main purpose in using a macroscopic model is to reduce considerably the number of unknowns to handle.
LibMultiScale is a parallel C++ framework for the multiscale coupling methods dedicated to material mechanics. This framework is designed with the form of a library providing an API, which makes possible the implementation of coupled simulations, and several clients with advanced configuration possibilities.
The coupling parts are provided by LibMultiScale while existing projects furnish the molecular dynamics and finite element continuum mechanics. In fact, the API provides C++ templated interfaces in order to reduce the cost of integration. For example, molecular dynamics softwares that have been integrated are Stamp (a code of the CEA) and Lammps (Sandia laboratories). Concerning continuum mechanics and finite elements, libMesh and Akantu have been plugged for use with LibMultiScale.
Recently LibMultiScale project has been fully reshaped and will be release on the LSMS web site (soon). This new release will include a functional user manual describing the possibilities of the configuration files. Together with AKANTU and LAMMPS they form a parallel, flexible and modern tool to address direct multiscale simulation issues.
Moreover, current developpements include the possibility of a three-way coupling between LAMMPS (MD), AKANTU (FEM) and ParaDiS (dislocation dynamics) to allow the 3D dynamical passing of dislocations from a MD part to its continuum surrondings. This is developped as the CADD3D project.
Finally the finite temperature condition is necessary to envision more realistic simulation of matter. Especially for the case of sliding contact, where large amounts of kinetic energy are released from the contacting surfaces, it is crucial to treat heat flows correctly and to prevent kinetic energy to be trapped in the molecular dynamics regions. We work on this problematic under the finite temperature project .