To calculate complex molecular-dynamics simulations efficiently, optimizations in areas like data layout and traversal patterns of the data managed by the simulation have a high significance. Especially, if the simulation is run on a GPU.

This thesis explores how algorithms and data structures that are used in the C++17 molecular-dynamics library AutoPas can be implemented with the Kokkos programming model. The focus is on what has to be considered for the data management in Kokkos, regarding data layout and traversal patterns. The created simulation then enables in-depth performance analyses on different computing platforms.