Constant-time one-shot testing of large-scale graph states
Fault-tolerant measurement-based quantum computation (MBQC) with recent progress on quantum technologies leads to a promising scalable platform for realizing quantum computation, conducted by preparing a large-scale graph state over many qubits and performing single-qubit measurements on the state. With fault-tolerant MBQC, even if the graph-state preparation suffers from errors occurring at an unknown physical error rate, we can suppress the effect of the errors. Verifying graph states is vital to test whether we can conduct MBQC as desired even with such errors. However, problematically, existing state-of-the-art protocols for graph-state verification by fidelity estimation have required measurements on many copies of the entire graph state and hence have been prohibitively costly in terms of the number of qubits and the runtime. We here construct an efficient alternative framework for testing graph states for fault-tolerant MBQC based on the theory of property testing. Our test protocol accepts with high probability when the physical error rate is small enough to make fault-tolerant MBQC feasible and rejects when the rate is above the threshold of fault-tolerant MBQC. The novelty of our protocol is that we use only a single copy of the $N$-qubit graph state and single-qubit Pauli measurements only on a constant-sized subset of the qubits; thus, the protocol has a constant runtime independently of $N$. Furthermore, we can immediately use the rest of the graph state for fault-tolerant MBQC if the protocol accepts. These results achieve a significant advantage over prior art for graph-state verification in the number of qubits and the total runtime. Consequently, our work offers a new route to a fast and practical framework for benchmarking large-scale quantum state preparation.