Speaker
Description
Quantum chemistry has become an important tool to analyze and predict the properties of molecules. Coupled cluster (CCSD(T)) is considered to be the gold standard as it gives accurate results and can be improved in a well-known manner, but is computationally expensive. If one wants to compute the properties of molecules containing heavy elements relativistic contributions need to be included and require a 2- or 4-component treatment for accurate results. This increases the amount of memory and floating point operations. Supercomputers provide such resources, but are heterogeneous systems with various memory spaces and processing units. An implementation of relativistic coupled cluster for such infrastructures will be presented, see also https://arxiv.org/abs/2103.08473.
