Quantum circuit architecture search for variational quantum algorithms

Abstract Variational quantum algorithms (VQAs) are expected to be a path advantages on noisy intermediate-scale devices. However, both empirical and theoretical results exhibit that the deployed ansatz heavily affects performance of VQAs such an with larger number gates enables stronger expressivity, while accumulated noise may render poor trainability. To maximally improve robustness trainability VQAs, here we devise resource runtime efficient scheme termed architecture search (QAS). In particular, given learning task, QAS automatically seeks near-optimal (i.e., circuit architecture) balance benefits side-effects brought by adding more achieve good performance. We implement numerical simulator real hardware, via IBM cloud, accomplish data classification chemistry tasks. problems studied, experimental show cannot only alleviate influence barren plateaus but also outperforms pre-selected ansatze.