This code postprocesses data obtained from MD simulations. This is most optimal for analyzing interface structure and mass transport at interfaces.

This code has been used in extracting information from MD trajectories of

1. Point defect structure and transport in Cu-Nb, Cu-V, Ag-V, Cu-Mo

2. Point defect structure and transport of MgO grain boundaries

This code has been used to identify interface structures for fcc-bcc metal interfaces without a reference state

Please see [some talks (speakerdeck)] (http://bit.ly/kedar-talks-speakerdeck) and [publications list (google scholar)] (http://bit.ly/Kedar-google-scholar-main)

Above activities are analogous to identifying features in AI problems. Unlike bulk properties of materials, interface properties of materials are not understood well -- so we don't know what "features" effect the performance of materials with materials.

Limitations of the state of the art: MD-like simulations used currently are still labor intensive. The data and information obtained by these methods is still very tightly dependent on the empericial potentials we use and it is hard to validate the results especially of those in nanomaterials because the physical phenomenon discovered in MD are of the time and length scale that are still not accessbile to experiments. This can change with development of new computational infrastructure and there is a push in this direction with the Materials Discovery Initiative by the White House.