Evaluating the flexibility of A* for mapping quantum circuits

Mapping quantum circuits to real quantum architectures (while keeping the respectively considered cost as small as possible) has become an important research task since it is required to execute algorithms on real devices. Since the underlying problem is NP-complete, several heuristic approaches have been proposed. Recently, approaches utilizing A* search to map quantum circuits to, e.g., Nearest Neighbor architectures or IBM QX architectures have gained substantial interest. However, their performance usually has only been evaluated in a rather narrow context, i.e., for single architectures and objectives only. In this work, we evaluate the flexibility of A* in the context of mapping quantum circuits to physical devices. To this end, we review the underlying concepts and show its flexibility with respect to the considered architecture. Furthermore, we demonstrate how easy such solutions can be adjusted towards optimizing different design objectives or cost metrics by providing a generalized and parameterizable cost function for the A* search that can also be easily extended to support future cost metrics.