Efficient and machine-learnable multi-qubit operations on a modular quantum processor with all-to-all reconfigurable coupling

Quantum algorithms on near-term NISQ processors are typically executed using shallow quantum circuits composed of one- and two-qubit gates. However, as circuit depth and gate number increase, this design paradigm becomes increasingly unreliable, ultimately limiting algorithmic complexity. An alternative approach is to investigate gates involving larger numbers of qubits. In previous work (X. Wu et al., arXiv:2407.20134 (2024)), we demonstrated a new architecture with user-selectable two-qubit interactions via a reconfigurable router used to connect pairs of qubits. Here, we extend this approach to enable programmable and efficient multi-qubit operations involving more than two qubits, with which we demonstrate faster preparation of multi-qubit entangled states with improved fidelities. We also successfully apply model-free reinforcement learning to the operation of multi-qubit entangling gates, including two-qubit controlled-Z and three-qubit controlled-swap gates, demonstrating the feasibility of engineering complex many-body quantum dynamics with our high-connectivity qubit coupling design. This promises new approaches to implementing complicated quantum algorithms and more practical quantum computing deployments.