Abstract

The usage of quantum annealers for solving optimisation problems has been of increasing interest in recent years, especially in the context of machine learning applications. Those real-life optimisation problems often require multi-body terms and higher connectivity, while most quantum annealing architectures only allow for two-body interactions and limited connectivity. In this talk we present recent work which aims to address those limitations. Firstly, we introduce an efficient circuit design for multi-body terms in a superconducting flux qubit annealer with native two-body interactions. Furthermore, it is shown how one can use those circuits to build a unit cell for a scalable quantum annealing architecture. Secondly, we use the theoretical construction underlying the circuit design to derive embedding schemes to map maximum satisfiability problems into the Chimera graph, which is the hardware graph of the current D-wave architecture.

Based on work done in collaboration with N. Chancellor, P Warburton, S. Benjamin, S. Roberts [arXiv:1603.09521, 1604.00651].