Variational quantum compiling with double Q-learning

Quantum compiling aims to construct a quantum circuit V by gates drawn from native gate alphabet, which is functionally equivalent the target unitary U. It crucial stage for running of algorithms on noisy intermediate-scale (NISQ) devices. However, space structure exploration enormous, resulting in requirement human expertise, hundreds experimentations or modifications existing circuits. In this paper, we propose variational (VQC) algorithm based reinforcement learning (RL), order automatically design VQC with no intervention. An agent trained sequentially select alphabet and qubits they act double Q-learning \epsilon-greedy strategy experience replay. At first, randomly explores number circuits different structures, then iteratively discovers structures higher performance task. Simulation results show that proposed method can make exact compilations less compared previous algorithms. reduce errors due decoherence process noise NISQ devices, enable especially complex be executed within coherence time.